fup: add cheddar as a syntax highlighter

web/fup/cmd/serve.go

@@ -9,12 +9,14 @@ import (
 	"fmt"
 	"log"
 	"net/http"
+	"os/exec"
 	"strings"
 
 	"github.com/google/safehtml"
 	"github.com/spf13/cobra"
 	"hg.lukegb.com/lukegb/depot/web/fup/fuphttp"
 	"hg.lukegb.com/lukegb/depot/web/fup/fupstatic"
+	"hg.lukegb.com/lukegb/depot/web/fup/minicheddar"
 )
 
 func init() {
@@ -24,13 +26,18 @@ func init() {
 	serveCmd.Flags().StringVar(&serveStaticRoot, "static-root", "/static/", "Root address from which static assets should be referenced.")
 	serveCmd.Flags().StringVarP(&serveBind, "listen", "l", ":8191", "Bind address for HTTP server.")
 	serveCmd.Flags().BoolVar(&serveDirectOnly, "direct-only", false, "If set, all file serving will be proxied, even if the backend supports signed URLs.")
+
+	serveCmd.Flags().StringVar(&serveCheddarPath, "cheddar-path", "cheddar", "Path to 'cheddar' binary to use for syntax highlighting. If it cannot be found, syntax highlighting and markdown rendering will be disabled.")
+	serveCmd.Flags().StringVar(&serveCheddarAddr, "cheddar-address", "", "If non-empty, will be used instead of attempting to spawn a copy of cheddar.")
 }
 
 var (
-	serveBind       string
-	serveRoot       string
-	serveStaticRoot string
-	serveDirectOnly bool
+	serveBind        string
+	serveRoot        string
+	serveStaticRoot  string
+	serveDirectOnly  bool
+	serveCheddarPath string
+	serveCheddarAddr string
 
 	serveCmd = &cobra.Command{
 		Use:   "serve",
@@ -44,6 +51,11 @@ var (
 			}
 
 			ctx := context.Background()
+			highlighter, err := serveCheddar(ctx)
+			if err != nil {
+				return fmt.Errorf("spawning cheddar syntax highlighter: %v", err)
+			}
+
 			cfg := &fuphttp.Config{
 				Templates:           fupstatic.Templates,
 				Static:              fupstatic.Static,
@@ -51,6 +63,7 @@ var (
 				AppRoot:             serveRoot,
 				StorageURL:          bucketURL,
 				RedirectToBlobstore: !serveDirectOnly,
+				Highlighter:         highlighter,
 			}
 			a, err := fuphttp.New(ctx, cfg)
 			if err != nil {
@@ -63,3 +76,17 @@ var (
 		},
 	}
 )
+
+func serveCheddar(ctx context.Context) (*minicheddar.Cheddar, error) {
+	if serveCheddarAddr != "" {
+		return minicheddar.Remote(serveCheddarAddr), nil
+	}
+
+	cpath, err := exec.LookPath(serveCheddarPath)
+	if err != nil {
+		log.Printf("couldn't find cheddar at %q; disabling syntax highlighting", serveCheddarPath)
+		return nil, nil
+	}
+
+	return minicheddar.Spawn(ctx, cpath)
+}

web/fup/fuphttp/fuphttp.go

@@ -27,6 +27,11 @@ const (
 	defaultRedirectExpiry = 5 * time.Minute
 )
 
+type Highlighter interface {
+	Markdown(ctx context.Context, text string) (safehtml.HTML, error)
+	Code(ctx context.Context, filename, theme, text string) (safehtml.HTML, error)
+}
+
 type Config struct {
 	Templates  fs.FS
 	Static     fs.FS
@@ -48,6 +53,10 @@ type Config struct {
 	// FilenameGenerator returns a new filename based on the provided prefix and extension.
 	FilenameGenerator fngen.FilenameGenerator
 
+	// Highlighter is used for syntax highlighting and Markdown rendering.
+	// If nil, then no syntax highlighting or Markdown rendering will be performed.
+	Highlighter Highlighter
+
 	// UseDirectDownload decides whether the "pretty" wrapped page or the direct download page is the most appropriate for a given set of parameters.
 	UseDirectDownload func(fileExtension string, mimeType string) bool
 }
@@ -59,6 +68,8 @@ type Application struct {
 
 	appRoot string
 
+	highlighter Highlighter
+
 	redirectToBlobstore bool
 	redirectExpiry      time.Duration
 

web/fup/minicheddar/minicheddar.go

@@ -0,0 +1,171 @@
+// SPDX-FileCopyrightText: 2021 Luke Granger-Brown <depot@lukegb.com>
+//
+// SPDX-License-Identifier: Apache-2.0
+
+package minicheddar
+
+import (
+	"bytes"
+	"context"
+	"encoding/json"
+	"errors"
+	"fmt"
+	"net"
+	"net/http"
+	"os/exec"
+	"time"
+
+	"github.com/google/safehtml"
+	"github.com/google/safehtml/uncheckedconversions"
+)
+
+const startupMax = 10 * time.Second
+
+type Cheddar struct {
+	remote string
+	stop   func()
+}
+
+type markdownReqResp struct {
+	Data string `json:"data"`
+}
+
+func (c *Cheddar) Markdown(ctx context.Context, text string) (safehtml.HTML, error) {
+	buf := new(bytes.Buffer)
+
+	if err := json.NewEncoder(buf).Encode(markdownReqResp{Data: text}); err != nil {
+		return safehtml.HTML{}, fmt.Errorf("encoding markdown rendering request: %w", err)
+	}
+
+	req, err := http.NewRequestWithContext(ctx, "POST", fmt.Sprintf("%s/markdown", c.remote), buf)
+	if err != nil {
+		return safehtml.HTML{}, fmt.Errorf("making markdown request object: %w", err)
+	}
+	req.Header.Set("Content-type", "application/json")
+
+	resp, err := http.DefaultClient.Do(req)
+	if err != nil {
+		return safehtml.HTML{}, err
+	}
+	defer resp.Body.Close()
+
+	if resp.StatusCode != http.StatusOK {
+		return safehtml.HTML{}, fmt.Errorf("markdown rendering request returned status %s; not OK", resp.StatusCode)
+	}
+
+	var respData markdownReqResp
+	if err := json.NewDecoder(resp.Body).Decode(&respData); err != nil {
+		return safehtml.HTML{}, fmt.Errorf("decoding markdown rendering response: %w", err)
+	}
+
+	return uncheckedconversions.HTMLFromStringKnownToSatisfyTypeContract(respData.Data), nil
+}
+
+type codeReq struct {
+	Filepath string `json:"filepath"`
+	Theme    string `json:"theme"`
+	Code     string `json:"code"`
+}
+
+type codeResp struct {
+	IsPlaintext bool   `json:"is_plaintext"`
+	Data        string `json:"data"`
+}
+
+func (c *Cheddar) Code(ctx context.Context, filename, theme, text string) (safehtml.HTML, error) {
+	buf := new(bytes.Buffer)
+
+	if err := json.NewEncoder(buf).Encode(codeReq{
+		Filepath: filename,
+		Theme:    theme,
+		Code:     text,
+	}); err != nil {
+		return safehtml.HTML{}, fmt.Errorf("encoding code rendering request: %w", err)
+	}
+
+	req, err := http.NewRequestWithContext(ctx, "POST", fmt.Sprintf("%s", c.remote), buf)
+	if err != nil {
+		return safehtml.HTML{}, fmt.Errorf("making code request object: %w", err)
+	}
+	req.Header.Set("Content-type", "application/json")
+
+	resp, err := http.DefaultClient.Do(req)
+	if err != nil {
+		return safehtml.HTML{}, err
+	}
+	defer resp.Body.Close()
+
+	if resp.StatusCode != http.StatusOK {
+		return safehtml.HTML{}, fmt.Errorf("code rendering request returned status %s; not OK", resp.StatusCode)
+	}
+
+	var respData codeResp
+	if err := json.NewDecoder(resp.Body).Decode(&respData); err != nil {
+		return safehtml.HTML{}, fmt.Errorf("decoding code rendering response: %w", err)
+	}
+
+	if respData.IsPlaintext {
+		return safehtml.HTMLEscaped(respData.Data), nil
+	}
+	return uncheckedconversions.HTMLFromStringKnownToSatisfyTypeContract(respData.Data), nil
+}
+
+func (c *Cheddar) healthCheck(ctx context.Context) error {
+	_, err := c.Markdown(ctx, "")
+	return err
+}
+
+func (c *Cheddar) Close() {
+	if c.stop != nil {
+		c.stop()
+	}
+}
+
+// Remote connects to an already running instance of Cheddar.
+func Remote(url string) *Cheddar {
+	return &Cheddar{
+		remote: url,
+	}
+}
+
+// Spawn spawns a new instance of Cheddar listening on some port.
+// This is potentially race-y, because it relies on binding, then unbinding and spawning Cheddar.
+// But... it should be good enough.
+func Spawn(ctx context.Context, cpath string) (*Cheddar, error) {
+	l, err := net.Listen("tcp", "[::1]:0")
+	if err != nil {
+		return nil, fmt.Errorf("finding a port for cheddar: %w", err)
+	}
+	addr := l.Addr().(*net.TCPAddr)
+	port := addr.Port
+	if err := l.Close(); err != nil {
+		return nil, fmt.Errorf("closing port-sentinel listener: %w", err)
+	}
+
+	ctx, cancel := context.WithCancel(context.Background())
+	cmd := exec.CommandContext(ctx, cpath, "--sourcegraph-server", "--listen", fmt.Sprintf("[::1]:%d", port))
+	if err := cmd.Start(); err != nil {
+		return nil, fmt.Errorf("launching cheddar: %w", err)
+	}
+	c := &Cheddar{
+		remote: fmt.Sprintf("http://[::1]:%d", port),
+		stop:   cancel,
+	}
+
+	startupCtx, startupCancel := context.WithTimeout(ctx, startupMax)
+	defer startupCancel()
+
+	// Give cheddar a bit to start up before we start healthchecking it.
+	time.Sleep(1 * time.Millisecond)
+	for {
+		if err := c.healthCheck(startupCtx); errors.Is(err, context.DeadlineExceeded) {
+			cancel() // Kill the subprocess.
+			return nil, err
+		} else if err == nil {
+			break
+		}
+		time.Sleep(10 * time.Millisecond)
+	}
+
+	return c, nil
+}